[Gnucash-changes] fix yet another intermediate-math overflow bug.

[Derek Atkins]

Linas,

This code looks like it could still overflow if either of the denom's
are negative.  Granted, denoms _shouldn't_ be negative, but someone
could theoretically create one that way.

-derek

Linas Vepstas <linas at cvs.gnucash.org> writes:

> +  if (a.denom == b.denom)
> +  {
> +    if(a.num == b.num) return 0;
> +    if(a.num > b.num) return 1;
>      return -1;
>    }
> +
> +  if  ((a.denom > 0) && (b.denom > 0))
> +  {
> +    /* Avoid overflows using 128-bit intermediate math */
> +    qofint128 l = mult128 (a.num, b.denom);
> +    qofint128 r = mult128 (b.num, a.denom);
> +    return cmp128 (l,r);
> +  }
> +
> +  aa = a.num * a.denom;
> +  bb = b.num * b.denom;
> +
> +  if(aa == bb) return 0;
> +  if(aa > bb) return 1;
> +  return -1;
>  }
>  
>  
> _______________________________________________
> gnucash-changes mailing list
> gnucash-changes at gnucash.org
> https://lists.gnucash.org/mailman/listinfo/gnucash-changes
>
>

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord at MIT.EDU                        PGP key available